Intel Loses Steam Thread

Jim Keller is out at Tesla and in at Intel. Really interested to see what he’ll be working on.
He sure likes to jump around, doesn't he.

Consistency and teambuilding seems difficult with a head dude which comes and goes constantly. :p
 
Jim Keller is out at Tesla and in at Intel. Really interested to see what he’ll be working on.
He's probably no longer under AMD's NDA and could work on either CPU or GPU product development at Intel.

Edit: A little more on the Intel move:
Fudzilla reports that Keller (pictured below, right) will work alongside Intel SVP Raja Koduri (pictured below, left) and will report to Group President and Chief Engineering Officer Dr Venkata (Murthy) Renduchintala, who in turn reports directly to Intel CEO Brian Krzanich.
...
It is understood from Fudzilla’s insider sources that “Keller will be working on the hardware engineering part while Koduri will be working on the IP blocks. Raja will make the IP blocks for CPU and GPU while Keller will make the SoC.” Via this teamwork it is reckoned that Intel’s shift from 14nm to 10nm will be given a fillip.
http://hexus.net/tech/news/industry/117605-amd-zen-cpu-architect-jim-keller-becomes-svp-intel/
 
Last edited:
That guy sure moves around a lot. Is this normal for people holding such positions?

For someone at his level, I think it makes sense and is reasonable. He’s working on high level architecture and moves on to a new challenge once his job is done. Obviously, lots of demand for his services. Sounds like his role at Tesla just wasn’t a good fit.
 
There are probably multiple reasons for moving around, although having a history in an in-demand and rarefied field such as high-end VLSI and architecture is a significant one. Some of his tenures went on for a number of years, and not all of his departures equate to his work being done. His first departure from AMD happened when the version of K8 he was architect for was cancelled, and in the case of Tesla I'm curious what job was done specifically. His position was the result of a promotion last June, which occurred due to a series of executives heading Tesla's autopilot project leaving. That seems like a short time to have everything wrapped up for self-driving Teslas.
If he is moving on to Intel, isn't it a touch late to significantly influence the 14nm to 10nm transition, unless it's delayed by another 2-3 years?
 
Wouldn't a new chip by Keller and Raja be at least 3 years out?
 
(a dramatized tiny story about Intel's manufacturing issues)
Yeah, I read just the other day that cannonlake/10nm has been delayed AGAIN until next year this time, and instead we'll get ANOTHER "new" architecture on 14nm from Intel; "whiskey lake".

...Presumably so named because at this point, that's what the company will need to drown their compounded sorrows...
 
Do note, what Intel calls 10nm is not what everybody else calls 10nm. Contacted gate pitch and minimum metal pitch of Intel's 10nm is equivalent to TSMC/Samsungs 7nm.

Intel's 14nm is roughly equivalent to everbody else's 10nm (Glofo excepted). They don't do Manhattan routing, so there is a density penalty there.

Cheers
 
Do note, what Intel calls 10nm is not what everybody else calls 10nm. Contacted gate pitch and minimum metal pitch of Intel's 10nm is equivalent to TSMC/Samsungs 7nm.

However TSMC and Samsung are either shipping or about to ship their 7nm chips to customers already, so Intel has lost quite a bit of their former manufacturing process advantage.
 
Isn't everyone going EUV on 7nm? (or was Intel doing it at 5nm instead?)

Personally I think at some point it will make more sense to go multi-layer CPU instead of going for finer processes, just like they did for flash.
How challenging (economical) this would be from a production point of view I'm not quite sure.
 
Personally I think at some point it will make more sense to go multi-layer CPU instead of going for finer processes, just like they did for flash.
How challenging (economical) this would be from a production point of view I'm not quite sure.

Samsung's 7 nm process would probably be equivalent to the troubled Intel 10 nm process (both in feature sizes and also in the fact that they'd coexist in overlapping time periods )

Intel won't do EUV for their 10 nm, as far as it's reported
 
Samsung's 7 nm process would probably be equivalent to the troubled Intel 10 nm process (both in feature sizes and also in the fact that they'd coexist in overlapping time periods )

Intel won't do EUV for their 10 nm, as far as it's reported

TSMC seems to be more successfull:
Apple supplier TSMC (Taiwan Semiconductor Manufacturing Company) has started production on the next-generation 7-nanometer A12 chips that will be used in the 2018 iPhone lineup, reports Bloomberg.

TSMC in late April announced that its 7-nanometer process node has entered into high volume manufacturing, but did not specify that it was working on the Apple A12 processors set to be built into the iPhones that are expected in September.

tsmc_logo_new.jpg

The new 7-nanometer chips will offer approximately 40 percent power and area benefit over the 10-nanometer process used for the A11 processors in the 2017 iPhones. As Bloomberg says, the chips will be smaller, faster, and more efficient.
 
Yes, but I rate TSMC's 7nm as less bold due to the lack of EUV.

There's nothing wrong with that ofc, history (and custumers) may very well validate TSMC's choices over their competition

(still don't get the need for TSMC's logo in your post though )
 
Back
Top